German Edition: DOI: 10.1002/ange.201908039Reaction Dynamics Very Important PaperInternational Edition: DOI: 10.1002/anie.201908039

Directed Gas-Phase Synthesis of Triafulvene under Single-CollisionConditionsAaron M. Thomas, Long Zhao, Chao He, Galiya R. Galimova, Alexander M. Mebel,* andRalf I. Kaiser*

Abstract: The triafulvene molecule (c-C4H4)—the simplestrepresentative of the fulvene family—has been synthesized forthe first time in the gas phase through the reaction of themethylidyne radical (CH) with methylacetylene (CH3CCH)and allene (H2CCCH2) under single-collision conditions. Theexperimental and computational data suggest triafulvene isformed by the barrierless cycloaddition of the methylidyneradical to the p-electron density of either C3H4 isomer followedby unimolecular decomposition through elimination of atomichydrogen from the CH3 or CH2 groups of the reactants. Thedipole moment of triafulvene of 1.90 D suggests that thismolecule could represent a critical tracer of microwave-inactive allene in cold molecular clouds, thus defining con-straints on the largely elusive hydrocarbon chemistry in low-temperature interstellar environments, such as that of theTaurus Molecular Cloud 1 (TMC-1).

Introduction

Since the pioneering discovery of pentafulvenes morethan 100 years ago by Thiele,[1,2] the fulvenes triafulvene (1),pentafulvene (2), heptafulvene (3), and nonafulvene (4 ;Scheme 1) have fascinated the physical organic, theoretical,and organometallic chemistry communities from the funda-mental viewpoints of electronic structure[3] and chemicalbonding[4] as prototypes of a class of exotic hydrocarbonsformally obtained by cross-conjugating one ring and a meth-ylidene moiety through an exocyclic carbon–carbon doublebond. The resulting molecules are planar with C2v pointgroups and 1A1 electronic ground states in the case of 1–3,whereas nonafulvene (4) adopts a twisted geometry (C2) witha 1A electronic ground state owing to inefficient conjugationof the exocyclic carbon–carbon double bond with the nine-

membered ring.[5,6] The unique reactivity associated with thecross-conjugated p-electronic structure has led to the broaduse of fulvenes in the synthesis of natural products, such asloganin and sarracenin,[7, 8] and in transition-metal complexes,such as {(h5-C5Me5)Ti(h6-C5H4 = C(pC7H7)2)}2(m2,h

1,h1-N2).[9, 10] Fulvenes also serve as fundamental molecular build-ing blocks in fluorescent aromatic hydrocarbons, including 9-(cycloheptatrienylidene)fluorene,[11] 9-(methylidenecyclopro-pabenzene)fluorene, and 5-(methylidenecyclopropabenze-ne)dibenzocycloheptatriene.[12]

Triafulvene (1, c-C4H4, methylenecyclopropene)—a ther-modynamically less stable (99 kJmol@1)[13] isomer of vinyl-acetylene (5, HCCC2H3) and butatriene (6, H2CCCCH2 ;Scheme 2)—has received particular attention as it resembles

the prototype of the fulvene family, defining the simplestcross-conjugated hydrocarbon carrying a methylidene func-tional group in conjugation with a single carbon–carbondouble bond within the cycloalkene. Although triafulvenesequipped with bulky phenyl substituents, such as 2,3-diphen-yl-4,4-dicyanotriafulvene (C18H8N2), are stable monomers atroom temperature,[14] triafulvene (1), first synthesized bydehalogenation of 2-bromomethylenecyclopropane, wasfound to polymerize at temperatures above 80 K.[15, 16] In

Scheme 1. Molecular structures and symmetries of triafulvene (1),pentafulvene (2), heptafulvene (3), and nonafulvene (4).

Scheme 2. Energetically low lying structural isomers of the C4H4

molecule. Enthalpies of formation (DfH(298)) are given in kJmol@1

relative to vinylacetylene.[13]

[*] A. M. Thomas, Dr. L. Zhao, Dr. C. He, Prof. Dr. R. I. KaiserDepartment of Chemistry, University of Hawai’i at ManoaHonolulu, HI 96822 (USA)E-mail: ralfk@hawaii.edu

G. R. Galimova, Prof. Dr. A. M. MebelDepartment of Chemistry and BiochemistryFlorida International UniversityMiami, FL 33199 (USA)E-mail: mebela@fiu.edu

G. R. GalimovaSamara National Research UniversitySamara 443086 (Russia)

Supporting information and the ORCID identification number(s) forthe author(s) of this article can be found under:https://doi.org/10.1002/anie.201908039.

AngewandteChemieResearch Articles

15488 T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495

a later study on triafulvene (1) synthesis, its head-to-tail C8H8

dimer 7 was successfully isolated at temperatures above198 K,[17] and its formation was attributed to the strong dipolemoment (1.90 D) associated with the dipolar resonancestructure of triafulvene (1, Scheme 3).[18] Despite evidence

for the formation of triafulvene (1) as a reactive intermediate,as determined by ultraviolet (UV; 193/248 nm) irradiation ofbutatriene (6) in 10 K argon matrices,[19] the directed synthesisand isolation of triafulvene (1) in the gas phase has eludedsynthetic chemists over the last decades. Considering thedifficulties in preparation, short lifetimes above 80 K, and thetendency for dimerization, free triafulvenes are one of theleast explored classes of organic molecules.

Herein, we offer an exceptional glance into the funda-mentally elusive chemistry of the cyclic triafulvene (1)molecule and account for its gas-phase synthesis undersingle-collision conditions through bimolecular reactions ofthe methylidyne radical (CH, X2P) with methylacetylene(CH3CCH, X1A1) and allene (H2CCCH2, X1A1) by utilizingthe crossed molecular beams approach. Exploration at themolecular level affords fundamental perception into theunderlying reaction mechanisms leading to the synthesis ofhighly reactive molecules, such as fulvenes. The reactions areconducted in the gas phase under single-collision conditionsso that the primary reaction products fly apart undisturbedand hence do not undergo successive reactions, such ashydrogen-atom-assisted isomerization. Since dimerization oftriafulvene (1) is precluded as well, crossed molecular beamexperiments enable the versatile and clean gas-phase syn-thesis of the simplest fulvene. This system is also appealingfrom the physical organic chemistry viewpoint, since triaful-vene (1) represents a benchmark, thus enabling the elucida-tion of the chemical reactivity, bond-breaking processes, andthe formation of cross-conjugated cyclic molecules fromacyclic precursors as an outcome of a single collision, which isnot possible by traditional synthetic chemistry routes. Thisapproach therefore provides a novel path to gain access toa hitherto rather obscure class of strained hydrocarbonmolecules: substituted triafulvenes.

Results

Laboratory Frame

The reactive scattering signal of reactions of the methyl-idyne radical (CH, X2P) with allene (H2CCCH2, X1A1) andmethylacetylene (CH3CCH, X1A1) was observed at mass-to-charge ratios (m/z) from 48 to 52. For each system, these time-

of-flight (TOF) spectra are superimposable after scaling andhence derive from a single reaction channel, namely, CH(13 amu) + C3H4 (40 amu)!C4H4 (52 amu) + H (1 amu),whereby m/z 52 corresponds to the singly ionized atomic-hydrogen-loss reaction product (C4H4

+), with ion counts at m/z 51–48 arising from dissociative electron impact ionization ofneutral C4H4 products. The data therefore suggest that theC4H4 isomers are formed through a methylidyne versusatomic-hydrogen exchange pathway in the reaction of themethylidyne radical with the C3H4 isomers allene andmethylacetylene. Under our experimental conditions, thesignal-to-noise ratio for the fragment ion at m/z 51 (C4H3

+)was better than that observed for the parent ion (C4H4

+; m/z 52) for both systems studied (see Figure S1 in the Support-ing Information); therefore, TOF spectra and the fulllaboratory angular distributions were extracted at m/z 51(see Figures S2 and S3). The resulting laboratory angulardistributions display maxima near the CM angles of (52.4:0.4)88, about which both distributions are nearly symmetricaland span at least 4088 in the laboratory frame (Figure 1). Thesymmetry suggests that the reaction proceeds through indi-rect scattering dynamics involving at least one C4H5 reactionintermediate that ultimately undergoes unimolecular decom-position by emitting atomic hydrogen. Notably, the laboratorydata for the methylidyne reaction with allene and methyl-acetylene depict nearly identical TOF spectra and laboratoryangular distributions, which may indicate that similar featureson the C4H5 potential-energy surface (PES), such as decom-posing intermediates, ultimately govern the reaction dynam-ics. The signal at m/z 51 could not be fit with molecular-hydrogen loss (2 amu) leading to a heavy C4H3 counterfragment (51 amu). These fits would result in TOF spectrathat are too fast and laboratory angular distributions that aretoo broad.

As the hydrogen atom can be ejected from the methyl-idyne and/or from the C3H4 reactant, we conducted crossedbeam experiments of methylidyne (CH) with[D4]methylacetylene (CD3CCD) and [D4]allene (D2CCCD2)to extract the position(s) of the atomic-hydrogen anddeuterium losses. In the CH/D2CCCD2 and CH/CD3CCDsystems, the hydrogen- (C4D4) and deuterium-loss (C4HD3)products are expected to depict ion counts of the parentmolecule at m/z 56 and 55, respectively. The product at m/z 56cannot fragment to m/z 55, but only to m/z 54. Therefore, theidentification of signals at m/z 56 and 55 would representexplicit evidence of the formation of the hydrogen- (C4D4)and deuterium-loss product (C4HD3), respectively. In bothsystems, reactive scattering signals were observed at m/z 55and 54, but not at m/z 56. These findings alone suggest theexistence of the atomic-deuterium-loss channel, with deute-rium ejected from the C3D4 reactants to form C4HD3 products(55 amu; see Figures S4–S7). Therefore, the results of the CH/D2CCCD2 system reveal unambiguously the formation ofC4HD3 isomers with deuterium emitted from one of the fourchemically equivalent deuterium atoms of the [D4]allenereactant; the methylidyne moiety stays intact. However,methylacetylene has two sets of chemically distinct hydrogenatoms at the methyl (CH3) and ethynyl (CCH) moieties. Toelucidate the position(s) from which the hydrogen atom is (or

Scheme 3. Dipolar resonance structure of triafulvene (1) leading to thehead-to-tail dimer 7.

AngewandteChemieResearch Articles

15489Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495 T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org

hydrogen atoms are) emitted, we carried out crossedbeam experiments of [D1]methylidyne (CD) with partiallydeuterated [D3]methylacetylene (CD3CCH) and[D1]methylacetylene (CH3CCD). The hydrogen-atom-losschannel would result in the formation of C4D4 and C4H2D2

and signals at m/z 56 and 54 (see Figure S5). The reactivescattering experiments at the center-of-mass (CM) anglesrevealed that the CD/CH3CCD system yields a hydrogen-lossproduct (C4H2D2); the ion counts at m/z 54 suggest that theatomic emission originates from the methyl (CH3) group ofthe methylacetylene reactant.

Center-of-Mass Frame

While the laboratory data provide compelling evidencefor the methylidyne versus atomic hydrogen exchange path-way with the hydrogen atom emitted from allene and themethyl group of methylacetylene, respectively, our goal wasto elucidate the underlying reaction mechanism(s) by expos-ing the identity of the isomer(s) formed. A forward-convo-lution fitting of the laboratory data yields CM translationalenergy P(ET) and CM angular T(q) flux distributions thatoffer a glimpse into the reaction dynamics (Figure 2). Withinour error limits, the best-fit CM functions achieved usinga product mass combination of C4H4 (52 amu) plus H (1 amu)were nearly identical for both systems. The derived P(ET)distributions reveal maximum translational energy releases(Emax) of (181: 12) and (178: 12) kJmol@1. Accounting for

Figure 1. a) Time-of-flight (TOF) spectra and b) laboratory angular distribution recorded at m/z 51 (C4H3+) for the reaction of the methylidyne

radical with allene at a collision energy of (18.6:0.3) kJmol@1. c) TOF spectra and d) laboratory angular distribution recorded at m/z 51 (C4H3+)

for the reaction of the methylidyne radical with methylacetylene at a collision energy of (18.7:0.2) kJmol@1. The circles represent the experimentaldata and the solid lines the best fits.

Figure 2. a) Center-of-mass translational energy P(ET) and b) angularT(q) flux distributions for the reaction of the methylidyne radical withallene (blue) and methylacetylene (red) forming C4H4 isomer(s) byatomic-hydrogen emission. Solid lines represent the best fit, whileshaded areas indicate the experimental error limits.

AngewandteChemieResearch Articles

15490 www.angewandte.org T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495

the energy conservation, the high-energy cutoff of the C4H4

products formed without internal excitation is best under-stood with the maximum energy Emax related to the collisionenergy (EC) plus the reaction energy (DrG) available forproduct translational degrees of freedom by Emax = EC@DrG.The reaction energies calculated for the CH/allene and CH/methylacetylene systems were (@162: 12) and (@159:12) kJ mol@1, respectively. Furthermore, the P(ET) distribu-tions peaked at 36 and 33 kJmol@1. This off-zero peakingsuggests that the hydrogen atom is lost through a repulsiveenergy release involving a tight exit transition state. Thisprocess can be best understood in terms of the reversereaction of a hydrogen atom reacting with a closed-shell C4H4

isomer to form a C4H5 intermediate, which involves anentrance barrier. Also, the average translational energies ofthe products were calculated to be (53: 4) and (51:4) kJ mol@1, that is, fractions of (29: 3)% of the maximumenergy release to form C4H4. These findings suggest indirectscattering dynamics involving C4H5 intermediates(s) thatundergo unimolecular decomposition through tight exittransition state(s) as characterized by a repulsive energyrelease and significant electron rearrangement to form thefinal C4H4 products.[20]

Finally, the T(q) distributions contain additional informa-tion on the scattering dynamics in revealing product flux overthe entire angular range; furthermore, these distributionsshow a pronounced forward–backward symmetry with a max-imum at 9088 (sideways scattered). The symmetry of the T(q)distributions suggests a decomposing C4H5 intermediate witha lifetime comparable to or exceeding its rotational period. Ina sense the reaction intermediate “lost” its memory of howthe reactants approached initially, and the intermediate thusloses a hydrogen atom in all directions. The sidewaysscattering in particular indicates strong geometrical con-straints on the decomposing intermediate with a preferencefor atomic-hydrogen loss perpendicular to the rotationalplane of the decomposing intermediate and hence nearlyparallel to the total angular momentum vector.[20, 21] Thus, forthe reverse reaction, the hydrogen atom adds perpendicularlyto the C4H4 product to allow maximum overlap of the 1sorbital of the hydrogen atom with the p-electron density ofthe C4H4 molecule.

Discussion

We now combine the aforementioned experimental find-ings with available data from electronic structure calculationsin an attempt to elucidate the underlying reaction dynamicsand the nature of the structural isomer(s) produced. Ribeiroand Mebel explored the C4H5 PES and revealed theformation of three distinct C4H4 isomers through the reac-tions of methylidyne with allene and methylacetylene.[22]

These three isomers are vinylacetylene (HCCC2H3 ; p1;DrG = (@269: 5) kJmol@1), 1,2,3-butatriene (H2CCCCH2 ;p2 ; DrG = (@239: 5) kJmol@1), and triafulvene (c-C4H4 ; p3 ;DrG = (@171: 5) kJmol@1), with overall reaction energiesgiven for the methylidyne/methylacetylene system. Thecrossed molecular beams approach has the unique advantage

of extracting the nature of the isomer(s) formed by comparingthe experimentally determined reaction energies ((@162:12) kJ mol@1; (@159: 12) kJ mol@1) with the computed reac-tion energies for distinct C4H4 isomers p1–p3 (Figure 3).[22]

These data are in strong agreement with the formation oftriafulvene in both systems; the formation of the vinylacety-lene (p1) or 1,2,3-butatriene (p2) isomers would increase theexperimental reaction exoergicity by 98 or 65 kJ mol@1, whichis clearly outside the error limits of the experiments.

By examining the PES, we can gain insight into thereaction dynamics leading to the formation of triafulvene(p3). The electronic structure of the methylidyne radical in its2P electronic ground state opens the possibility for two typesof barrierless entrance channels through addition to the p-electron density at carbon–carbon double or triple bonds orinsertion into C@H or C@C s bonds. Key reaction pathwaysleading to triafulvene (p3) are highlighted in Figure 4. In theCH/CH3CCH system, barrierless cycloaddition of the meth-ylidyne radical to the acetylene moiety forms cyclic collisioncomplexes i1 and i2, whose molecular geometries aredistinguished by the location of the unpaired electron.Intermediate i1 possesses a planar carbon skeleton but onlyC1 symmetry due to the nonsymmetric positioning of thehydrogen atoms in the CH3 group. The radical is essentiallydelocalized on the two sp2-hybridized carbon atoms in thethree-membered ring carrying the C@H groups and linked byan ordinary C@C bond. Notably, the rotation of the methylgroup is nearly free, and hence a Cs symmetric structure withthe plane of symmetry containing all four carbon atoms isalmost degenerate with the C1 local minimum i1. In i2 theradical center resides on the carbon atom carrying the methylgroup, and the C@H groups in the ring are linked by a doubleC=C bond, thus resulting in a Cs symmetric structure with themirror plane perpendicular to the three-membered ring. Bothintermediates can interconvert readily with a barrier of only20 kJ mol@1. Alternatively, intermediate i1 can undergo a si-multaneous hydrogen migration–ring opening to yield inter-mediate i8, which in turn isomerizes through hydrogenmigration–ring closure to i9. The unimolecular decompositionof both i2 and i9 releases a hydrogen atom, thereby establish-ing the cross-conjugated p system in the heavy coproduct:triafulvene (p3). In the CH/H2CCCH2 system, the methyl-idyne radical can add to the carbon–carbon double bond toyield the collision complex i9.

The theoretically predicted geometries of the exit tran-sition states involved in the i2!p3 + H and i9!p3 + Hpathways depict the hydrogen atom departing at angles of 8.9and 3088, respectively, with respect to the total angularmomentum vector. These geometries are consistent with thebest-fit T(q) distributions, whose shapes are suggestive ofa sideways-scattered reaction product, whereby the C4H4 fluxin the methylacetylene and allene reaction is highest at 9088,with 62 and 75 %, respectively, of the maximum product fluxappearing near the poles (088 and 18088). Importantly, the T(q)for the CH/H2CCCH2 reaction is isotropic within the errorlimits of the experimental data and may account for therelatively wide-angled departure of the hydrogen atom for thei9!p3 + H pathway.

AngewandteChemieResearch Articles

15491Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495 T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org

Most importantly, our experiments exploiting (partially)deuterated reactants support the aforementioned findings.First, the experiments along with the extracted reaction

mechanisms reveal that the hydrogenatom is emitted from the methyl group(CH/CH3CCH) and from the CH2 moiety(CH/H2CCCH2). In the case of theCD/H2CCCH2 system, this process is ra-tionalized by the addition of the[D1]methylidyne radical to the carbon–carbon double bond of allene to yieldintermediate d1-i9, which ejects atomichydrogen to give [D1]triafulvene (d1-p3 ;Figure 5). For CD/CH3CCD, the initialaddition of [D1]methylidyne forms d2-i1/d2-i2 ; the latter can eject atomic hydrogento give [D2]triafulvene (d2-p3). In principle,d2-i1 could isomerize to d2-i8, whichmay rearrange to d2-i9 and emit atomicdeuterium to give [D1]triafulvene (d1-p3 ;C4H3D; 53 amu; Figure 5). However, theabsence of atomic-hydrogen-loss signalsfrom methylidyne (CH) or the acetylenicend of CH3CCH, as determined fromthe CH/CD3CCD and CD/CD3CCH reac-

tions (see Figure S5), suggests that intermediate i8 doesnot play a significant role in the reaction dynamicsof the methylidyne–methylacetylene system and that

Figure 3. Potential energy surface for the reaction of the methylidyne radical (CH) with allene (H2CCCH2) and methylacetylene (CH3CCH).Energies calculated at the CCSD(T)/CBS level are shown in kJmol@1 and are relative to the energy of the methylidyne–methylacetylene reactants.

Figure 4. Potential energy surface for the reaction of the methylidyne (CH) radical with allene(H2CCCH2) and methylacetylene (CH3CCH) leading to triafulvene (p3). Energies calculated atthe CCSD(T)/CBS level are shown in kJ mol@1 and are relative to the energy of themethylidyne–methylacetylene reactants.

AngewandteChemieResearch Articles

15492 www.angewandte.org T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495

the mechanism involves the unimolecular decomposition ofd2-i2.

Previous experimental studies under “bulk” conditionsessentially led to the conclusion that the reactions ofmethylidyne with allene and methylacetylene proceeded byan addition or insertion mechanism that ultimately yields low-lying C4H4 isomers alongside hydrogen atoms. Daugey et al.proposed that methylidyne radicals undergo fast reactionswith allene and methylacetylene at rates of few 4 X10@10 cm@3 molecule@1 s@1 (77–170 K);[23] products could notbe identified. Goulay et al. probed the methylidyne reactionwith allene and methylacetylene using vacuum ultraviolet(VUV) photoionization to photoionize the products. Theirstudy sampled vinylacetylene (p1), 1,2,3-butatriene (p2), andtriafulvene (p3)/cyclobutadiene, with the latter only appear-ing in the methylacetylene system.[24] Although the contribu-

tion of the cyclic C4H4 isomerstriafulvene and/or cyclobutadienecould not be disentangled undertheir experimental conditions, therecent ab initio and statistical anal-ysis by Ribeiro and Mebel foundan overall agreement with theirbulk-phase (4 torr) product distri-bution with the conclusion that thecyclic isomer was most likely tri-afulvene.[22] As compared to thesingle-collision conditions affordedby the present crossed molecularbeams study, reaction intermedi-ates undergo a few thousand colli-sions with the buffer gas at rela-tively high pressures of 4 torr;therefore, the internal energy ofany intermediates can be divertedto the buffer gas, thus providinga different outcome of the reactionowing to distinct internal energiesof the reaction intermediates thatultimately influence the measuredproduct distribution.

We also carried out energy-dependent Rice–Ramsperger–Kassel–Marcus (RRKM) calcula-tions of product branching ratiosunder single-collision condi-tions[25, 26] using the potential en-ergy surface and molecular param-eters computed by Ribeiro andMebel (see Table S2 in the Sup-porting Information).[22] At the ex-perimental collision energy of19 kJ mol@1, methylidyne additionto the triple bond in methylacety-lene is statistically predicted toresult in the production of about67% triafulvene, 18% vinylacety-lene, and 14 % butatriene. Alter-natively, insertion of methylidyne

into a C(sp3)@H bond and the single C@C bond should nearlyexclusively produce vinylacetylene, whereas the insertion intothe C(sp)@H bond should form nearly equal amounts ofvinylacetylene and butatriene at about 47 and 53 %, respec-tively. Since the triple-bond addition is dynamically preferred,our present experimental results for the methylidyne plusmethylacetylene reaction qualitatively agree with the statis-tical picture showing triafulvene as the dominant reactionproduct. However, since the RRKM approach based onchemically accurate ab initio calculations of relative energiesand molecular parameters has been shown to closely repro-duce experimental product branching ratios for numerouschemical reactions when the system explored exhibits stat-istical behavior, the non-observation of vinylacetylene andbutatriene in experiment indicates deviation from the stat-istical RRKM behavior. The contrast between the RRKM

Figure 5. Bimolecular reaction of the [D1]methylidyne (CD) radical with [D1]methylacetylene (CH3CCD)and allene (H2CCCH2) leading to triafulvene (p3) isotopologues through hydrogen- and/or deute-rium-atom emission. Energies are given for the fully hydrogenated species; energies of the (partially)deuterated species differ by a few kJmol@1 at most.

AngewandteChemieResearch Articles

15493Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495 T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org

and experimental results is much more dramatic for themethylidyne–allene system, for which the calculations predictthat the dynamically favorable methylidyne addition to thedouble bond should almost solely form vinylacetylene,whereas the insertion into a C@H bond should give mostlyacetylene plus vinyl radicals with vinylacetylene as a minorproduct. Negligible amounts of triafulvene are statisticallyformed in both cases, and thus the methylidyne–allene systemunder single-collision conditions clearly demonstrates non-RRKM behavior of this system.

Conclusion

In conclusion, our study enabled the first unambiguousidentification of the triafulvene (p3 ; c-C4H4) isomer asa reaction product of the gas-phase reaction of methylidyne(CH) radicals with allene (H2CCCH2) and methylacetylene(CH3CCH) under single-collision conditions. Triafulvene wasformed by the barrierless cycloaddition of the methylidyneradical to the p-electron density, thus resulting in rovibra-tionally excited C4H5 intermediates that undergo unimolec-ular decomposition through hydrogen-atom emission. Sys-tematic analysis of the crossed molecular beams datacombined with electronic structure calculations suggests thatisomerization on the C4H5 PES proceeds most likely throughan addition/elimination mechanism. Beyond fundamentallyabstract considerations, the reaction mechanisms uncoveredin this study are most relevant to the combustion andastrochemical regimes. With respect to the combustionscenario, the C4H5 reaction intermediates, for example, couldreact with acetylene (C2H2) to eventually form C6H6 isomers,such as fulvene and ultimately benzene,[27, 28] linking them tothe sooting process via polycyclic aromatic hydrocarbons(PAHs) through mass-growth protocols, such as the hydro-gen-abstraction C2H2-addition (HACA) mechanism.[29, 30] Tri-afulvene, if formed, could undergo collision-induced isomer-ization to the thermodynamically favored vinylacetylene (p1;HCCC2H3)—a key component for PAH formation andgrowth in a complex reaction scheme with applicability atthe opposing extremes of hydrocarbon combustion andinterstellar space,[31,32] where, for example, the CH andCH3CCH molecules are known to exist in high abundance.Despite this insight, our study under single-collision con-ditions highlights that the reactivity of nascent triafulvene andthe effects of its cross-conjugated p system on the radical-driven chemistry of high-temperature combustion systems areessentially unknown.

Experimental Section

The methylidyne (CH; X2P) gas-phase reactions with methyl-acetylene (CH3CCH; X1A1) and allene (H2CCCH2 ; X1A1) along withtheir fully and partially deuterated isotopologues were conductedunder single-collision conditions using the crossed molecular beamsmethod.[33] A pulsed supersonic beam of methylidyne radicals wasproduced by photodissociation of bromoform (CHBr3, AldrichChemistry, + 99%), first cooled to 283 K then diluted in helium(99.9999%; AirGas), using the focused 248 nm output of an excimer

laser (Coherent; CompEx 110) 1 mm downstream of the nozzle.[34]

The methylidyne radicals were velocity-selected by a chopper wheelbefore crossing perpendicularly with a supersonic beam of puremethylacetylene (Organic Technologies, 99% +) or allene(H2CCCH2 ; Organic Technologies, 99% +) with resulting collisionenergies of 19 kJmol@1 (see Table S1); typical rotational temperaturesof the methylidyne radical beam of (14: 1) K were achieved.[33]

Where necessary, experiments with (partially) deuterated reactantswere carried out utilizing [D1]methylidyne (CD), produced analo-gously to the methylidyne radical by exploiting CDBr3 (AldrichChemistry, 99.5% +) as a precursor, [D3]methylacetylene (CD3CCH;CDN Isotopes, 99% +), [D4]methylacetylene (CD3CCD; CDN Iso-topes, 99 % +), [D1]methylacetylene (CH3CCD; CDN Isotopes,99% +), and [D4]allene (D2CCCD2 ; CDN Isotopes, 98% +) asappropriate.

The neutral reaction products were ionized by electron impact(80 eV) and then filtered according to their m/z ratio using a quadru-pole mass spectrometer. The detection equipment was housed withina differentially pumped and rotatable chamber that allowed thecollection of product velocity distributions by measuring angularlyresolved time-of-flight (TOF) spectra in the plane of the reactantbeams. The laboratory data were analyzed by a forward-convolutionroutine[35, 36] that relies on user-defined center-of-mass (CM) transla-tional energy P(ET) and angular T(q) flux distributions that are variediteratively until a best fit of the experimental dataset is obtained.These functions comprise the reactive differential cross section I(q,u), which is taken to be separable into its CM scattering angle q andCM velocity u components, I(u, q)&P(u) X T(q). Errors of the P(ET)and T(q) functions were determined within the 1s limits of thecorresponding laboratory angular distribution and beam parameters(beam spreads, beam velocities) while maintaining a good fit of thelaboratory TOF spectra.

Computational Details

We employed RRKM theory[37–39] to calculate energy-dependentrate constants of all unimolecular reaction steps on the C4H5 potentialenergy surface (PES) accessed after the initial association of themethylidyne radical with allene and methylacetylene. The PES itselfand molecular parameters of all the species (reactants, products,intermediates, and transition states) were computed earlier andreported in a previous publication.[22] In this study, internal-energy-dependent rate constants were assessed within the harmonic approx-imation employing our in-house code,[40] which automatically pro-cesses GAUSSIAN09 log files to evaluate numbers of states fortransition states and densities of states for local minima using thedirect count method. The calculations were carried out at the zero-pressure limit corresponding to crossed molecular beam conditions.RRKM rate constants were then used to evaluate product branchingratios by solving first-order kinetic equations within steady-stateapproximation.[25, 40] This ab initio/RRKM approach has been shownto closely reproduce experimental product branching ratios when thesystem exhibits statistical behavior.[32, 41–43]

Acknowledgements

This research was supported by the U.S. Department ofEnergy, Basic Energy Sciences DE-FG02-03ER15411 andDE-FG02-04ER15570 to the University of Hawaii and toFlorida International University, respectively.

AngewandteChemieResearch Articles

15494 www.angewandte.org T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495

Conflict of interest

The authors declare no conflict of interest.

Keywords: gas-phase reactions · radicals · reaction dynamics ·strained molecules · triafulvene

How to cite: Angew. Chem. Int. Ed. 2019, 58, 15488–15495Angew. Chem. 2019, 131, 15634–15641

[1] J. Thiele, Ber. Dtsch. Chem. Ges. 1900, 33, 666 – 673.[2] J. Thiele, H. Balhorn, Justus Liebigs Ann. Chem. 1906, 348, 1 –

15.[3] S. Belz, T. Grohmann, M. Leibscher, J. Chem. Phys. 2009, 131,

034305.[4] H. Mçllerstedt, M. C. Piqueras, R. Crespo, H. Ottosson, J. Am.

Chem. Soc. 2004, 126, 13938 – 13939.[5] E. Kleinpeter, A. Fettke, Tetrahedron Lett. 2008, 49, 2776 – 2781.[6] M. Neuenschwander, Helv. Chim. Acta 2015, 98, 763 – 784.[7] H.-M. Tai, M.-Y. Chang, A.-Y. Lee, N.-C. Chang, J. Org. Chem.

1999, 64, 659 – 662.[8] P. Preethalayam, K. S. Krishnan, S. Thulasi, S. S. Chand, J.

Joseph, V. Nair, F. Jaroschik, K. V. Radhakrishnan, Chem. Rev.2017, 117, 3930 – 3989.

[9] A. Scherer, K. Kollak, A. Lgtzen, M. Friedemann, D. Haase, W.Saak, R. Beckhaus, Eur. J. Inorg. Chem. 2005, 1003 – 1010.

[10] R. Beckhaus, Coord. Chem. Rev. 2018, 376, 467 – 477.[11] B. Hu, P. Lu, Y. Wang, New J. Chem. 2013, 37, 1645 – 1653.[12] Y. Apeloig, R. Boese, B. Halton, A. H. Maulitz, J. Am. Chem.

Soc. 1998, 120, 10147 – 10153.[13] D. Cremer, E. Kraka, H. Joo, J. A. Stearns, T. S. Zwier, Phys.

Chem. Chem. Phys. 2006, 8, 5304 – 5316.[14] H. L. Ammon, J. Am. Chem. Soc. 1973, 95, 7093 – 7101.[15] W. E. Billups, L. J. Lin, E. W. Casserly, J. Am. Chem. Soc. 1984,

106, 3698 – 3699.[16] S. W. Staley, T. D. Norden, J. Am. Chem. Soc. 1984, 106, 3699 –

3700.[17] W. E. Billups, C. Gesenberg, R. Cole, Tetrahedron Lett. 1997, 38,

1115 – 1116.[18] T. D. Norden, S. W. Staley, W. H. Taylor, M. D. Harmony, J. Am.

Chem. Soc. 1986, 108, 7912 – 7918.[19] R. Wrobel, W. Sander, D. Cremer, E. Kraka, J. Phys. Chem. A

2000, 104, 3819 – 3825.[20] R. D. Levine, Molecular Reaction Dynamics, Cambridge Uni-

versity Press, Cambridge, UK, 2005.[21] W. B. Miller, S. A. Safron, D. R. Herschbach, Discuss. Faraday

Soc. 1967, 44, 108 – 122.[22] J. M. Ribeiro, A. M. Mebel, Phys. Chem. Chem. Phys. 2017, 19,

14543 – 14554.

[23] N. Daugey, P. Caubet, B. Retail, M. Costes, A. Bergeat, G.Dorthe, Phys. Chem. Chem. Phys. 2005, 7, 2921 – 2927.

[24] F. Goulay, A. J. Trevitt, G. Meloni, T. M. Selby, D. L. Osborn,C. A. Taatjes, L. Vereecken, S. R. Leone, J. Am. Chem. Soc.2009, 131, 993 – 1005.

[25] V. V. Kislov, T. L. Nguyen, A. M. Mebel, S. H. Lin, S. C. Smith, J.Chem. Phys. 2004, 120, 7008 – 7017.

[26] A. N. Morozov, A. M. Mebel, R. I. Kaiser, J. Phys. Chem. A2018, 122, 4920 – 4934.

[27] H. R. Hrodmarsson, J. C. Loison, U. Jacovella, D. M. P. Holland,S. Boy8-P8ronne, B. Gans, G. A. Garcia, L. Nahon, S. T. Pratt, J.Phys. Chem. A 2019, 123, 1521 – 1528.

[28] C. Huang, B. Yang, F. Zhang, G. Tian, Combust. Flame 2018, 198,334 – 341.

[29] M. Frenklach, Phys. Chem. Chem. Phys. 2002, 4, 2028 – 2037.[30] M. Frenklach, H. Wang, Symp. Combust. [Proc.] 1991, 23, 1559 –

1566.[31] L. Zhao, R. I. Kaiser, B. Xu, U. Ablikim, M. Ahmed, M. M.

Evseev, E. K. Bashkirov, V. N. Azyazov, A. M. Mebel, Nat.Astron. 2018, 2, 973 – 979.

[32] R. I. Kaiser, D. S. N. Parker, A. M. Mebel, Annu. Rev. Phys.Chem. 2015, 66, 43 – 67.

[33] R. I. Kaiser, P. Maksyutenko, C. Ennis, F. Zhang, X. Gu, S. P.Krishtal, A. M. Mebel, O. Kostko, M. Ahmed, Faraday Discuss.2010, 147, 429 – 478.

[34] P. Maksyutenko, F. Zhang, X. Gu, R. I. Kaiser, Phys. Chem.Chem. Phys. 2011, 13, 240 – 252.

[35] P. S. Weiss, PhD Dissertation thesis, University of California(Berkeley, CA), 1986.

[36] M. F. Vernon, PhD Dissertation thesis, University of California(Berkeley, CA), 1983.

[37] P. J. Robinson, K. A. Holbrook, Unimolecular Reactions, Wiley,New York, 1972.

[38] H. Eyring, S. H. Lin, S. M. Lin, Basic Chemical Kinetics, JohnWiley and Sons, Inc., New York, NY, 1980.

[39] J. Steinfield, J. Francisco, W. Hase, Chemical Kinetics andDynamics, Prentice Hall, Englewood Cliffs, 1982.

[40] C. He, L. Zhao, A. M. Thomas, A. N. Morozov, A. M. Mebel,R. I. Kaiser, J. Phys. Chem. A 2019, 123, 5446 – 5462.

[41] R. I. Kaiser, A. M. Mebel, Int. Rev. Phys. Chem. 2002, 21, 307 –356.

[42] X. Gu, R. I. Kaiser, A. M. Mebel, ChemPhysChem 2008, 9, 350 –369.

[43] A. M. Mebel, R. I. Kaiser, Int. Rev. Phys. Chem. 2015, 34, 461 –514.

Manuscript received: June 28, 2019Accepted manuscript online: July 31, 2019Version of record online: September 3, 2019

AngewandteChemieResearch Articles

15495Angew. Chem. Int. Ed. 2019, 58, 15488 – 15495 T 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org